What Is Happening with My Connections? Why Are They Closed?
summary
Diagnoses unexpected connection closures with concrete fixes and demos: uses Spring WebClient on Reactor Netty, captures packets with Wireshark and correlates requests via Spring observability trace ID. Shows a load balancer silently dropping idle connections (4 minute idle timeout) and three client fixes—connection-pool max-idle+eviction, aggressive TCP keepalive (start 30s, probe every 10s), and delayed keepalive that detects dead peers after 8 probes. Also demos: 2s server request timeout, 1s client response timeout, and a 5-connection rate limit.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 0:48Common causes of closures
- 5:04Troubleshooting connection issues
- 6:52Preventing connection problems
- 8:57Demo: Load balancer idle timeout
- 18:53Demo: Server request timeout
- 22:11Demo: Rate limiting on server
- 24:36Demo: Client response timeout
- 27:34Summary and resilience strategies
- 28:15Q&A session
related talks
54mHow to build Reactive Server in 50 minutes
26mconnect() - why you so slow?!
31mDebugging the Routing Tier
01h 52m100% Fast and Resilient Microservice with RSocket
21mMaking Envoy Resilient to Sudden Increases in Load
06mLivin' on the Edge (of your cluster)
05mJump-starting back-office connections
24mResponsiveness Under Working Conditions